Final Project

Airbnb Data Visualization

library(dplyr)
library(tidyr)
library(tidytext)
library(ggplot2)
library(wordcloud)
library(ggwordcloud)
library(plotly)
library(lubridate)
library(leaflet)
library(tidygeocoder)
library(scales)
library(hexbin)
library(viridis)

Read in the data & initial check

# read the data
airbnb <- read.csv("Airbnb_Open_Data.csv")

# factor all the characters
airbnb <- airbnb |> mutate(across(where(is.character), as.factor))

# checking missing values
colSums(is.na(airbnb))

                            id                           NAME 
                             0                              0 
                       host.id         host_identity_verified 
                             0                              0 
                     host.name            neighbourhood.group 
                             0                              0 
                 neighbourhood                            lat 
                             0                              8 
                          long                        country 
                             8                              0 
                  country.code               instant_bookable 
                             0                            105 
           cancellation_policy                      room.type 
                             0                              0 
             Construction.year                          price 
                           214                              0 
                   service.fee                 minimum.nights 
                             0                            409 
             number.of.reviews                    last.review 
                           183                              0 
             reviews.per.month             review.rate.number 
                         15879                            326 
calculated.host.listings.count               availability.365 
                           319                            448 
                   house_rules                        license 
                             0                              0 

sum(duplicated(airbnb)) # there are 541 duplicated data

[1] 541

airbnb <- airbnb[-which(duplicated(airbnb)), ] # remove the duplicated rows
dim(airbnb) # there are 102058 rows and 26 variables

[1] 102058     26

Tidy the dataset

levels(airbnb$neighbourhood.group)

[1] ""              "Bronx"         "brookln"       "Brooklyn"     
[5] "manhatan"      "Manhattan"     "Queens"        "Staten Island"

# rename the levels of neighbourhood group
airbnb <- airbnb |>
  mutate(
    neighbourhood.group = case_when(
      neighbourhood.group %in% c("brookln") ~ "Brooklyn",
      neighbourhood.group %in% c("manhatan") ~ "Manhattan",
      TRUE ~ neighbourhood.group
    ),
    neighbourhood.group = factor(neighbourhood.group)
  )
levels(airbnb$neighbourhood.group)

[1] ""              "Bronx"         "Brooklyn"      "Manhattan"    
[5] "Queens"        "Staten Island"

# The neighbourhoods with neighborhood group empty
airbnb |>
  filter(neighbourhood.group == "") |>
  pull(neighbourhood)

 [1] Washington Heights Clinton Hill       East Village       Upper East Side   
 [5] Woodside           Williamsburg       Bushwick           Prospect Heights  
 [9] East Village       Williamsburg       Clinton Hill       Chelsea           
[13] Prospect Heights   East Harlem        Bushwick           Eastchester       
[17] Williamsburg       Harlem             Chinatown          Williamsburg      
[21] Queens Village     Harlem             Williamsburg       Bedford-Stuyvesant
[25] East Village       East Harlem        Harlem             Bushwick          
[29] Upper West Side   
225 Levels:  Allerton Arden Heights Arrochar Arverne Astoria ... Woodside

# group those area into the a county
lookup <- tibble::tribble(
  ~neighbourhood,       ~borough,
  "Washington Heights", "Manhattan",
  "Clinton Hill",       "Brooklyn",
  "East Village",       "Manhattan",
  "Upper East Side",    "Manhattan",
  "Woodside",           "Queens",
  "Williamsburg",       "Brooklyn",
  "Bushwick",           "Brooklyn",
  "Prospect Heights",   "Brooklyn",
  "Chelsea",            "Manhattan",
  "East Harlem",        "Manhattan",
  "Eastchester",        "Bronx",
  "Harlem",             "Manhattan",
  "Chinatown",          "Manhattan",
  "Queens Village",     "Queens",
  "Bedford-Stuyvesant", "Brooklyn",
  "Upper West Side",    "Manhattan"
)

airbnb <- airbnb |>
  left_join(lookup, by = "neighbourhood") |>
  mutate(
    neighbourhood.group = if_else(
      neighbourhood.group == "",
      borough,                 # fill with correct borough
      neighbourhood.group      # keep existing value
    ),
    neighbourhood.group = as.factor(neighbourhood.group)
  ) |>
  select(-borough)
levels(airbnb$neighbourhood.group)

[1] "Bronx"         "Brooklyn"      "Manhattan"     "Queens"       
[5] "Staten Island"

# remove the $ in the price variable
airbnb$price <- gsub("\\$", "", airbnb$price)
airbnb$price <-  as.numeric(airbnb$price)

Visualizations

Count of Listings by Neighborhood Group

neighbour_list <- ggplot(airbnb, aes(x = neighbourhood.group))+
  geom_bar(fill = "darkorchid", width = 0.6)+
  labs(
    title = "Number of Listings per Neighbourhood",
    x = "Neighbourhood group",
    y = "Count"
  )+
  theme_minimal()

ggplotly(neighbour_list)

Average Price by Room Type

price_per_room <- airbnb |>
  group_by(room.type) |>
  summarise(avg_price= mean(price, na.rm = T)) 

price_roomtype <- ggplot(price_per_room, aes(x = room.type, y = avg_price)) +
  geom_bar(stat = "identity", fill = "turquoise", width = 0.5)+
  geom_text(aes(label = round(avg_price,2)))+
  scale_y_continuous(labels = dollar_format())+
  labs(
    title = "Average Price by Room Type",
    x = "Room Type",
    y = "Average Price($)"
  ) +
  theme_minimal()

ggplotly(price_roomtype)

Listings by Price and Room Type

list_room_price <- ggplot(airbnb, aes(x = room.type, y = neighbourhood.group, fill= price))+
  geom_tile()+
  scale_fill_viridis(option = "plasma", 
                     direction = -1,
                     )+
  labs(
    title = "Prices by Room Type & Neighborhood",
    x = "Room Type", 
    y = "Neighborhood Group",
    fill = "Price ($)"
  )

ggplotly(list_room_price)

Map

nyc <- map_data("county") |> filter(region == "new york")
label_df <- airbnb |> 
  group_by(neighbourhood.group)  |> 
  summarise(
    long = mean(long, na.rm = TRUE),
    lat  = mean(lat, na.rm = TRUE)
  )

label_df <- airbnb |> 
  group_by(neighbourhood.group)  |> 
  summarise(
    long = mean(long, na.rm = TRUE),
    lat  = mean(lat, na.rm = TRUE),
    n_listings = n(),
    avg_price = mean(price, na.rm = TRUE)
  )

map <- ggplot() +
  geom_polygon(data = nyc, aes(x = long, y = lat, group = group),
               fill = "white", color = "gray") +
  coord_quickmap(xlim = c(-74.3, -73.7), ylim = c(40.49,40.9)) +
  geom_hex(data = airbnb, aes(x = long, y = lat), bins = 1000) +
  scale_fill_gradient(low = "skyblue", high = "red3") +
  geom_text(data = label_df,
            aes(x = long, y = lat, 
                label = neighbourhood.group, 
                text = paste(
        "Total listings:", n_listings,
        "\n Avg price: $", round(avg_price, 1)
      )),
            color = "black",
            size = 3) +
  labs(title = "Airbnb House Map") +
  theme_classic()

map

ggplotly(map, tooltip = "text")

Word Cloud

airbnb <- airbnb |> 
  mutate(house_rules = as.character(house_rules))

# word frequency table
word_freq <- airbnb |> 
  filter(house_rules != "") |> # remove missing house rules
  unnest_tokens(bigram, house_rules, token = "ngrams", n = 2) |> 
  # lowercasing all; token the sentences into bigram(2 words for a phase)
  separate(bigram, c("word1", "word2"), sep = " ")  |> # separate the bigram
  filter(
    !is.na(word1), !is.na(word2), # remove NA
    !word1 %in% stop_words$word | !word2 %in% stop_words$word 
    # remove the bigram with both words are stop words
  )  |> 
  unite(bigram, word1, word2, sep = " ")  |>  # recombine the bigram
  count(bigram, sort = TRUE)
head(word_freq, 10)

          bigram     n
1     no smoking 26067
2        no pets 11741
3       check in 10188
4      the house  8626
5  the apartment  7754
6      check out  7395
7   the building  6653
8     no parties  6311
9  be respectful  6142
10    smoking no  6137

word_freq |> filter(n > 1300) |> 
  ggplot(aes(label = bigram, size = n, color = n)) +
  geom_text_wordcloud() +
  scale_size_area(max_size = 20) +
  scale_color_gradient(low = "#0072B2", high = "#E69F00") +
    theme_minimal()